Steps to reproduce:

/etc/zabbix/zabbix_agent2.conf:

...Plugins.Mysql.Sessions.MySQLMonitor.Uri=tcp://localhost:3306
Plugins.Mysql.Sessions.MySQLMonitor.User=zbx_monitor
Plugins.Mysql.Sessions.MySQLMonitor.Password=xxx

Test MySQL availability:

zabbix_get -s 127.0.0.1 -k "mysql.ping[MySQLMonitor]"

result: 1

Then stop mysql/mariadb:

systemctl stop mariadb

Check again:

zabbix_get -s 127.0.0.1 -k "mysql.ping[MySQLMonitor]"

result: 0

Now start mariadb again:

systemctl start mariadb

Check again:

zabbix_get -s 127.0.0.1 -k "mysql.ping[MySQLMonitor]"

result: still 0

Restart Agent:

systemctl restart zabbix-agent2

Check again:

zabbix_get -s 127.0.0.1 -k "mysql.ping[MySQLMonitor]"

result: 1

Expected:
Even when using a named session, the session should be tried to restart automatically after a failure in order not to present wrong status to Zabbix.